| Benchmark | Gemini 2.5 Flash | Qwen3.7 Plus Preview |
|---|---|---|
| AA Agentic Index | 18.8 | 20.8 |
| AA Intelligence | 17.7 | 39.4 |
| AA-LCR | 65.7 | 69 |
| AA-Omniscience | -29.8 | 2.4 |
| arena_text_factuality | 1437 | 1454 |
| arena_vision | 1214 | 1263 |
| Artificial Analysis Coding Index | 22.2 | 55.9 |
| critpt | 1.4 | 9.1 |
| gdpval | 11.9 | 22.2 |
| GPQA Diamond | 82.8 | 90.3 |
| HLE | 12.1 | 35.6 |
| IFBench | 50.3 | 79.1 |
| MMMU-Pro | 69.1 | 80.5 |
| OmniScience Accuracy | 26.1 | 22.5 |
| OmniScience Non-Hallucination | 24.7 | 74.5 |
| scicode | 35.9 | 51.3 |
| SWE-bench Verified | 60.4 | 77.7 |
| Terminal-Bench Hard | 13.6 | 47 |
| τ²-Bench Telecom (AA run) | 31.6 | 93 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.